Laboratory measurements, paleontological data, and well -logs are often used to conduct mineralogical and chemical analyses to classify rock samples. Employing digital intelligence techniques may enhance the accuracy of classification predictions while simultaneously speeding up the whole classification process. We aim to develop a comprehensive approach for categorizing igneous rock types based on their global geochemical characteristics. Our strategy integrates advanced clustering, classification, data mining, and statistical methods employing worldwide geochemical data set of -25,000 points from 15 igneous rock types. In this pioneering study, we employed hierarchical clustering, linear projection analysis, and multidimensional scaling to determine the frequency distribution and oxide content of igneous rock types globally. The study included eight classifiers: Logistic Regression (LR), Gradient Boosting (GB), Random Forest (RF), K -nearest Neighbors (KNN), Support Vector Machine (SVM), Artificial Neural Network (ANN), and two ensemble -based classifier models, EN1 and EN -2. EN -1 consisted of LR, GB, and RF aggregates, whereas EN -2 comprised the predictions of all ML models used in our study. The accuracy of EN -2 was 99.2 %, EN -1 achieved 98 %, while ANN yielded 98.2 %. EN 2 provided the best performance with highest initial curve for longest time on the receiver operating characteristic (ROC) curve. Based on the ranking features, SiO 2 was deemed most important followed by K 2 O and Na 2 O. Our findings indicate that the use of ensemble models enhances the accuracy and reliability of predictions by effectively capturing diverse patterns and correlations within the data. Consequently, this leads to more precise results in rock typing globally.

Throughout the last few years, there has been a significant increase in demand for high temperature capacitors due to the rising need for electronics in harsh environments, including aerospace, automotive, and oil and gas industries. With the continued growth of electronics in these environments, there is a need for the development of new materials and manufacturing techniques to drive advances in high temperature capacitor technology. Additive manufacturing is one promising approach for producing electronics that can withstand high temperatures, leading to improved performance and reliability in a wide range of applications. The present work describes the fabrication and testing of MIM (metal-insulator-metal) capacitors at high temperatures using gold as the conductor material. The substrates used in this work is 3D printed Lithoz 350D 99.8% alumina. The work details the fabrication process and evaluates the relative permittivity of high-temperature dielectric material. The adhesion strength at the interfaces is examined before and after aging for up to 250hours at temperatures as high as 750°C. The leakage current is measured for 100hours at varying temperatures and the insulation resistance is calculated. Furthermore, the capacitance is monitored during aging at temperatures up to 700°C and frequencies as high as 1MHz. Finally, the capacitance is measured at room temperature before and after subjecting the capacitors to 100 cycles of thermal cycling at three different temperature ranges. The adhesion between all interfaces of the MIM capacitors is found to remain high even after aging at 700°C, leakage currents are minimal and stable, and the capacitance remained very stable during and after aging and after thermal cycling.

Reservoir characterization through seismic data analysis is essential for exploration and production in the petroleum industry. However, seismic-to-well tie discrepancies, limited availability of high-quality well data, and resolution constraints pose a reliability challenge. While previous studies offer valuable insights, they still struggle to achieve high-resolution predictions in a complex geologically environment given high reliance on well data. This study integrates synthetic data-driven techniques with real data, including convolutional neural networks (CNN) and transfer learning, to improve seismic reservoir characterization. We utilize nearby well statistics and a rock physics model (RPM) to simulate pseudo wells representing various geological scenarios. Synthetic seismic gathers are generated from these pseudo wells, which are based on RPM and local well control, to train the CNN. Transfer learning is then applied to adapt the CNN to better distinguish between real and synthetic data, enhancing reservoir predictions. A comparative analysis of P-impedance predictions from three methodologies: theory-driven Pre-Stack-Seismic-Inversion (TDSI), Deep-Neural-Network (DNN), and our CNN approach, showed that CNN achieved nearly 97

Abstract With the current development of the 5G infrastructure, there presents a unique opportunity for the deployment of battery-less mmWave reflect-array-based sensors. These fully-passive devices benefit from having a larger detectability than alternative battery-less solutions to create self-monitoring megastructures. The presented ‘smart’ skin sensor uses a Van-Atta array design enabling ubiquitous local strain monitoring for the structural health monitoring of composite materials featuring wide interrogation angles. Proof-of-concept prototypes of these ‘smart’ skin millimeter-wave identification tags, that can be mounted on or embedded within common materials used in wind turbine blades, present a highly-detectable radar cross-section of − 33.75 dBsm and − 35.00 dBsm for mounted and embedded sensors respectively. Both sensors display a minimum resolution of 202 $$\upmu $$ μ -strain even at 40 $$^{\circ }$$ ∘ off-axis enabling interrogation of the fully-passive sensor at oblique angles of incidence. When interrogated from a proof-of-concept reader, the fully-passive, sticker-like mmID enables local strain monitoring of both carbon fiber and glass fiber composite materials. The sensors display a repeatable and recoverable response over 0–3000 $$\upmu $$ μ -strain and a sensitivity of 7.55 kHz/ $$\upmu $$ μ -strain and 7.92 kHz/ $$\upmu $$ μ -strain for mounted and embedded sensors, respectively. Thus, the presented 5G-enabled battery-less sensor presents massive potential for the development of ubiquitous Digital Twinning of composite materials in future smart cities architectures.

Lithofacies identification plays a pivotal role in understanding reservoir heterogeneity and optimizing production in tight sandstone reservoirs. In this study, we propose a novel supervised workflow aimed at accurately predicting lithofacies in complex and heterogeneous reservoirs with intercalated facies. The objectives of this study are to utilize advanced clustering techniques for facies identification and to evaluate the performance of various classification models for lithofacies prediction. Our methodology involves a two-information criteria clustering approach, revealing six distinct lithofacies and offering an unbiased alternative to conventional manual methods. Subsequently, Gaussian Process Classification (GPC), Support Vector Machine (SVM), Artificial Neural Network (ANN), and Random Forest (RF) models are employed for lithofacies prediction. Results indicate that GPC outperforms other models in lithofacies identification, with SVM and ANN following suit, while RF exhibits comparatively lower performance. Validated against a testing dataset, the GPC model demonstrates accurate lithofacies prediction, supported by synchronization measures for synthetic log prediction. Furthermore, the integration of predicted lithofacies into acoustic impedance versus velocity ratio cross-plots enables the generation of 2D probability density functions. These functions, in conjunction with depth data, are then utilized to predict synthetic gamma-ray log responses using a neural network approach. The predicted gamma-ray logs exhibit strong agreement with measured data (R2 = 0.978) and closely match average log trends. Additionally, inverted impedance and velocity ratio volumes are employed for lithofacies classification, resulting in a facies prediction volume that correlates well with lithofacies classification at well sites, even in the absence of core data. This study provides a novel methodological framework for reservoir characterization in the petroleum industry.

The most crucial elements in the oil and gas sector are predicting subsurface lithofacies utilizing geophysical logs for reservoir characterization and sweet spot assessment procedures. Nevertheless, accurately predicting payable lithofacies in a complex heterogeneous geological setting, such as the lower goru formation, poses considerable difficulty because conventional methods fall short in delivering highly accurate outcomes. Hence, this research proposes an advanced cost and time-saving data intelligence strategy using multiple classifiers to predict lithofacies with maximum accuracy that will aid in sweet spot evaluation in oil and gas fields globally. Geophysical log data of five wells from a mature gas field were used. The targeted reservoir formation was classified into seven facies types. We evaluated the performance of seven different models: support vector machine (SVM), K-nearest neighbors (KNN), random forest (RF), decision tree (DTr), naive Bayes (NB), adaptive boosting (AB), and ensemble (an integrated SVM, KNN, RF, and DTr classifier). RF and ensemble classifiers predicted the lithofacies with accuracies of 97.5 and 97.3

Abstract The oil and gas industry relies on accurately predicting profitable clusters in subsurface formations for geophysical reservoir analysis. It is challenging to predict payable clusters in complicated geological settings like the Lower Indus Basin, Pakistan. In complex, high-dimensional heterogeneous geological settings, traditional statistical methods seldom provide correct results. Therefore, this paper introduces a robust unsupervised AI strategy designed to identify and classify profitable zones using self-organizing maps (SOM) and K-means clustering techniques. Results of SOM and K-means clustering provided the reservoir potentials of six depositional facies types (MBSD, DCSD, MBSMD, SSiCL, SMDFM, MBSh) based on cluster distributions. The depositional facies MBSD and DCSD exhibited high similarity and achieved a maximum effective porosity (PHIE) value of ≥ 15%, indicating good reservoir rock typing (RRT) features. The density-based spatial clustering of applications with noise (DBSCAN) showed minimum outliers through meta cluster attributes and confirmed the reliability of the generated cluster results. Shapley Additive Explanations (SHAP) model identified PHIE as the most significant parameter and was beneficial in identifying payable and non-payable clustering zones. Additionally, this strategy highlights the importance of unsupervised AI in managing profitable cluster distribution across various geological formations, going beyond simple reservoir characterization.

While there is a connection between petrophysical logs and reservoir porosity, finding analytical solutions for this relationship is still difficult. This paper presents a novel approach for forecasting porosity and lithofacies by using a convolutional neural network (CNN) model in conjunction with a bi-directional long short-term memory (BLSTM) network. The BLSTM network uses a self-organizing map (SOM) technique to form connections between input and destination data. The SOM is used to organize depth intervals with similar facies into four separate clusters, each exhibiting internal consistency in petrophysical parameters. The CNN is responsible for extracting spatial characteristics, while the BLSTM network gathers comprehensive spatiotemporal components, guaranteeing that the model accurately represents the spatiotemporal aspects of log data. The accuracy of the model was verified by analyzing simulation logging data. The findings indicate that the BLSTM network model successfully recovers significant characteristics from logging data, resulting in improved estimate accuracy. In addition, Facies-01 has lower gamma ray levels in comparison to other facies. Facies-01 is also suggestive of pristine sandstone formations, which are greatly sought as reservoir rocks. The BLSTM network model is effective in predicting physical characteristics of reservoirs, offering a new method for precise reservoir characterization parameter prediction.

Lake Fuxian is one of the deepest tectonic plateau freshwater lakes in the southeastern Tibetan Plateau, China. However, questions such as how old the lake is, how deep the total sedimentary thickness sequences are, and what landscape of the lake basin settings and geological structures are unknown. Here, based on fifteen seismic reflection profiles, we applied seismic facies and seismic sequence stratigraphic analyses to interpret the lake sequences. The results of the seismic response reveal that the maximum thickness of the sedimentation is ca. 1238 m and lies toward the NNE region of the lake basin on the L10-2 survey line. Lake sediments can be categorized into five seismic sequences and six seismic horizons. The oldest clinoforms in the deepest sequence (Sq-5) show that the depositional center was shifted to 19 km from the NNE region to the SSW modern location and was 930 m lower than the current lake floor. Multiple and complex tectonic activities strongly impacted on the lake basin, and a series of normal faults created an overall crustal extensional regime, resulting in the formation of many horst and graben structures.

The demand for advanced, high-performance radio frequency (RF) systems continues to rise in a wide range of applications, and embedded packaging is becoming a critical enabler for high-density electronics. Traditional RF packaging technologies are manufactured using various wafer fabrication lines, have various input/output (IO) pad metallization types, and/or are designed for probe station testing. As a result, it is often challenging to integrate these devices into a more complex module on a single RF substrate in a high-fidelity, low-parasitic way, which hinders the demonstration of next-generation communications architectures and concepts. In this work, we developed a fully additive manufacturing process for RF die packaging based on 3D printing for quick and flexible prototyping of embedded electronic modules and for improvements in size, weight, power, and cost. First, alumina 3D printing was used to produce a matrix for RF chip embedding and RF transmission line printing. Process conditions were optimized to fabricate small square pockets with lateral dimensions of 1 mm x 1 mm and depth of 150 microns. In addition, through plate holes with diameters as small as 300 microns and aspect ratio of ~2:1 were produced. Secondly, printed electronics methods like micro-dispensing (μDS) and aerosol jet printing (AJP) were employed to fabricate dielectric and conductive components. Custom RF test dies were manufactured and were pick-and-placed inside the pockets previously made in the alumina matrix. A conductive adhesive was micro-dispensed for die attachment, and PDS was also used to build a dielectric ramp bridging the die and the matrix. In the subsequent steps, AJP was used to print coplanar waveguide (CPW) transmission lines on the alumina surface and interconnects between those and the metalized pads on the test die. Printed components and RF die units were tested for mechanical and electrical performance and also for robustness and reliability under thermal cycling (-55°C to 125°C, 100 cycles) and aging (85°C, 85% relative humidity), showing excellent results. For conductive die-attach material, average adhesion strength was 13.5 Kgf before and 16.2 Kgf or 21.5 Kgf after aging or thermal cycling, respectively. Conductive vias in alumina presented a similar improvement in performance, with maximum DC resistance being 0.5 Ohms before and 0.35 Ohms after thermal cycling. CPW RF transmission lines with different geometries were simulated via electromagnetic (EM) modeling and a few variations with different pad sizes were printed for validation. Ink conductivity and 3D printed alumina roughness were used in the simulations, and good agreement between measured (0.22-0.29 dB/mm) and modeled (0.22-0.24 dB/mm) signal attenuation loss was observed. The integration of printed CPW lines with custom-manufactured microstrip lines on embedded silicon dies was also performed. This embedded unit exhibited low RF losses of 0.274 dB/mm at a frequency of 30 GHz. As conclusion, the combination of alumina 3D printing and printed electronic methods enabled highly customizable designs and easier prototyping for embedded RF circuits, as well as a path toward more complex RF multi-chip modules.

The Mahu sag slope area, which holds significance as an oil and gas resource, still have some underexplored regions because of structural mismatches, presenting a potential challenge to be properly addressed. To resolve, this study conducts a comprehensive investigation concerning structural characteristics, fault combinations, favorable reservoir distribution, reservoir control factors, and oil-water distribution characteristics within the Triassic Baikouquan formation, evaluating the impact of depositional environments and sedimentary dynamics on reservoir quality. For this purpose, constrained sparse spike inversion and seismic waveform indication inversion were employed to comparatively evaluate oil and gas reservoirs, further integrating petrophysical and geological data with geological modeling to enhance accuracy in complex structural geology and enable high-precision reservoir prediction. The findings elucidated the distribution range of the Baikouquan formation and the location of oil reservoir sand bodies, as exemplified by well B and identified potential hydrocarbon traps, offering valuable insights into reservoir performance. It demonstrated comparatively reliable effects and considerable predictability power of seismic waveform indication inversion. These outcomes provide a strong foundation for future evaluations and multi-layer system deployment in the region by serving as a novel valuable framework for subsequent development activities not only in the Mahu sag but also in similar regions.

Geoscientists now identify coal layers using conventional well logs. Coal layer identification is the main technical difficulty in coalbed methane exploration and development. This research uses advanced quantile–quantile plot, self-organizing maps (SOM), k-means clustering, t-distributed stochastic neighbor embedding (t-SNE) and qualitative log curve assessment through three wells (X4, X5, X6) in complex geological formation to distinguish coal from tight sand and shale. Also, we identify the reservoir rock typing (RRT), gas-bearing and non-gas bearing potential zones. Results showed gamma-ray and resistivity logs are not reliable tools for coal identification. Further, coal layers highlighted high acoustic (AC) and neutron porosity (CNL), low density (DEN), low photoelectric, and low porosity values as compared to tight sand and shale. While, tight sand highlighted 5–10% porosity values. The SOM and clustering assessment provided the evidence of good-quality RRT for tight sand facies, whereas other clusters related to shale and coal showed poor-quality RRT. A t-SNE algorithm accurately distinguished coal and was used to make CNL and DEN plot that showed the presence of low-rank bituminous coal rank in study area. The presented strategy through conventional logs shall provide help to comprehend coal-tight sand lithofacies units for future mining.

The Shangdan suture zone (SDZ) in the Qinling orogenic belt (QOB) is a key to understanding the East Asia tectonic evolution. The SDZ gives information about convergent processes between the North China Block (NCB) and South China Block (SCB). In the Late Mesozoic, several shear zones evolved along the SDZ boundary that helps us comprehend the collisional deformation between the NCB and SCB, which was neglected in previous studies. These shear zones play an essential role in the tectonic evolution of the East Asia continents. This study focuses on the deformation and geochronology of two shear zones distributed along the SDZ, identified in the Shaliangzi and Maanqiao areas. The shear sense indicators and kinematic vorticity numbers (0.54–0.90) suggest these shear zones have sinistral shear and sub-simple shear deformation kinematics. The quartz’s dynamic recrystallization and c-axis fabric analysis in the Maanqiao shear zone (MSZ) revealed that the MSZ experienced deformation under green-schist facies conditions at ∼400–500 °C. The Shaliangzi shear zone deformed under amphibolite facies at ∼500–700 °C. The 40Ar/39Ar (muscovite-biotite) dating of samples provided a plateau age of 121–123 Ma. Together with previously published data, our results concluded that QOB was dominated by compressional tectonics during the Late Early Cretaceous. Moreover, we suggested that the Siberian Block moved back to the south and Lhasa-Qiantang-Indochina Block to the north, which promoted intra-continental compressional tectonics.

Meshing of conventional PCB grounds refers to a process in which certain ground planes appear as copper lattices; regular openings are placed at regular intervals. The need for ground meshing for rigid PCBs has become minor with the development of micro-etching approaches. However, for flexible hybrid electronics, meshed grounds could offer some benefits, such as being unsusceptible to bending and conserving material and time. In this paper, we study how meshed ground planes affect the RF performance of straight microstrip lines. Simulations show meshed grounds with more than 50% filling produce good RF performance. When the filling percentage of the meshed grounds is less than 50%, ripples in the insertion loss start to appear. To confirm the simulation results, dispensing and aerosol jet printing systems were used to fabricate silver ink microstrip lines on PET substrates with different meshed ground patterns. The experimental measurement confirmed the simulation results. Bend testing was carried out to investigate the impact of mesh grounds on the RF performance of the microstrip lines after bending. The results demonstrate that the less-filled meshed ground samples are less susceptible to bending. As an extension of the work, we developed an empirical model to modify the microstrip line’s width to smoothen the insertion loss ripples while maintaining the bending superiority. To experimentally validate the model, predictions from this empirical model were used to fabricate and measure some samples.

Reservoir characterization is a vital task within the oil and gas industry, with the identification of lithofacies in subsurface formations being a fundamental aspect of this process. However, lithofacies identification in complex geological environments with high dimensions, such as the Lower Indus Basin in Pakistan, poses a notable challenge, especially when dealing with limited data. To address this issue, we propose four common data-driven machine learning approaches: multi-resolution graph-based clustering (MRGC), artificial neural networks (ANN), K-nearest neighbors (KNN), and self-organizing map (SOM). We utilized these proposed approaches to assess their performance in scenarios with varying core sample availability, specifically evaluating their effectiveness in identifying lithofacies within the Lower Goru formation of the middle Indus Basin. The study reveals that in scenarios with a limited number of core samples, MRGC is the preferred choice, while KNN or MRGC is more suitable for larger datasets. The results demonstrate the superior performance of MRGC and KNN in lithofacies identification within the specified geological environment, with SOM following closely behind, and ANN exhibiting comparatively lower efficacy. The accurate identification of lithofacies from the selected model is complemented by the application of the truncated Gaussian simulation method for facies modeling. Comparative results confirm the excellent agreement between the model identification of lithofacies from well logs and electro-facies obtained from the truncated Gaussian simulation electro-facies volume. This study highlights the crucial role of selecting the right machine learning approach for precise lithofacies identification and modeling in complex geological environments. The comparative analysis provides practitioners in the petroleum industry with insights into the strengths and limitations of each method, enhancing existing knowledge. In conclusion, this research emphasizes the significance of comprehensive research and method selection for advancing lithofacies identification in diverse formations or study areas, ultimately benefiting the broader field of subsurface characterization in the petroleum industry.

Direct writing methods created a revolution in the electronic industry due to their lower cost, fast processing, and lower wasted material. Microstrip line is an important electronic component that transfers the signal and the foundation for the communication between multiple components in any circuit board. Therefore, studying its electromechanical behavior against thermal and mechanical stresses is necessary for real-life applications. In this research, novel microstrip lines were printed on "polyethylene terephthalate" (PET) and "polyimide" (PI) flexible substrates using an aerosol jet printer (AJP) and dispensing system (DS). An advanced posttreatment technique was used to enhance the conductivity of the microstrip lines printed on low glass transition flexible PET substrate. Different microstrip line designs with various mesh ground planes were tested under mild and harsh mechanical bending and different environmental conditions and their losses were characterized. The results showed that the photonic curing enhanced the microstrip lines conductivity by 65% compared to the convectional curing. The in situ resistance measurements during harsh bending demonstrated conclusively that the robustness of the printed microstrip lines increased as the filling percentage of the ground plane became lower. The aging at 85 C-degrees/85% RH had a significantly stronger effect on the microstrip lines conductivity compared to the aging at 85 C-degrees without humidity due to the changes in the printed ink's microstructure and the increment in ionic conductivity. Thermal aging led to a reduction in the microstrip line's ductility and the cracking became easier in the microstructure of the printed films. The resistance of a sample aged at 85 C-degrees increased by 81.7% after 10000 bending cycles compared to only 20.5% for sample without thermal aging. Such findings provide important guidelines for those designing flexible hybrid electronics and for manufacturers who seek the assurance of these technologies for both maturing and reliable products.

In the relentless pursuit of expanding the boundaries of what is achievable under extreme temperature conditions, the precise measurement of RF signals becomes crucial. The ability to capture and analyze RF data in environments with high temperatures not only improves operational efficiency and reliability but also opens up new avenues for scientific investigation. This research introduces a innovative advancement in the field of high-temperature electronics: high-temperature SiO 2 cables equipped with edge launch connectors. These connectors represent a significant advancement in high-temperature electronics, as they are specifically engineered to withstand elevated temperatures, boasting an impressive resilience of up to 600°C. This exceptional temperature tolerance makes them invaluable assets in industries where electronic components are exposed to extreme heat conditions. To verify their performance, the RF characteristics of Coplanar Waveguide (CPW) lines and Metal-Insulator-Metal (MIM) capacitors were evaluated using the innovative SiO 2 cables and edge launch connectors. Remarkably, these connectors demonstrated outstanding performance, maintaining both their structural integrity and RF functionality even after being subjected to temperatures of up to 600°C for three complete cycles, with no signs of degradation observed. This outcome underscores the durability and reliability of these state-of-the-art connectors, positioning them as indispensable instruments for high-temperature RF measurements.

The development of innovative, miniaturized, and low-cost Radio Frequency Identification (RFID) tags for application in asset monitoring, counterfeit prevention, or personnel tracking requires advancements in materials, fabrication processes, and packaging. Typical RFID tags can be circumvented by tampering, cloning, or spoofing; however, by adding security features to the tag, nefarious actions such as these can be mitigated. Toward this objective, this paper presents the design and fabrication of an Ultra High Frequency (UHF) RFID tag through flexible hybrid electronics (FHE) materials and processes for authentication and anti-tamper /anti-counterfeit applications. The presented UHF RFID tag consists of a passive RF chip and dipole antenna with embedded hardware and software security features. The tag was fabricated using a hybrid of manufacturing techniques including, conventional photolithography and additive aerosol jet printing. The design, materials selection, processing, and tailored FHE fabrication processes, led to achieving a system-level functional RFID tag with a read distance of up to 15 in (0.381 m). The dependency of the read distance on the host surface was studied by attaching tags to different materials including surfaces with various dielectric constants and thicknesses. The performance of the tags was evaluated under realistic use conditions by performing thermal cycling, bending, and wearability tests. The RFID tag’s resistance to different tamper attack vectors (vulnerability assessment) is demonstrated. Overall, the demonstrated UHF RFID tag opens new opportunities for the development of flexible, lightweight, and low-cost RFID tags that leverage FHE fabrication techniques and materials for authentication and anti-tamper applications.

Lake Fuxian is a tectonic lake located on the Yunnan–Guizhou Plateau in southwest China. It is the deepest freshwater tectonic lake in the Yunnan Plateau. The present study focused on examining the structural changes, faulting patterns, and their influence on fault subsidence in the Lake Fuxian basin. Seismic interpretation showed uplift in the SSW area and subsidence in the NNE region. Subsidence is more pronounced on the northern survey lines, where the sedimentary strata had a maximum sedimentation of 1200 m. The seismic interpretation findings showed a horst block in the southern basin and a graben block in the northern half of the basin. L-14 demonstrated the steeper with maximum throw and parallel character of normal faults and provided the evidence of crustal extensional regime. Thirteen main faults were identified by fault modeling in the lake basin. The analysis of fault characteristics revealed that faults in the northern basin are characterized by greater depth, steeper angles, maximum displacement, and are actively moving owing to low resistance and negative asperity values, and poor edge detection values. Faults in the southern basin have an opposite character to those in the northern basin. Major faults in the northern lake basin have a stronger influence of fault subsidence compared to faults in the center and southern lake basins. Overall, the lake Fuxian basin showed horst-graben structure with parallel normal faulting with a crustal extensional regime.

This study investigates the application of machine learning techniques—specifically convolutional neural networks, multilayer perceptrons and cascaded forward neural networks —to understand the wettability of the CO2/brine/rock system, a critical factor in carbon dioxide (CO2) capture, utilization, and storage in deep saline aquifers. Understanding wettability is essential for improving the efficacy of CO2 storage. The study incorporates variables such as salinity, mineral types, measurement methods, pressure, and temperature into the machine learning models. Using a dataset of 876 samples from existing literature, the proposed models were trained and optimized using the Adam optimizer, Levenberg-Marquardt algorithm, and particle swarm optimization respectively.The performance of these models was evaluated through plot analysis, statistical indicators, and the Taylor diagram, demonstrating a high level of accuracy compared to experimental data. The specifically convolutional neural networks model showed exceptional accuracy in predicting CO2 wettability in brine, with a root mean square error of 0.9612 and coefficient of determination value of 0.9982. The minimal presence of outliers in the specifically convolutional neural networks model further confirms its robustness.This research highlights the effectiveness of deep learning in modeling complex wettability behaviors in CO2-brine-mineral systems, offering substantial insights for enhancing carbon dioxide (CO2) capture, utilization, and storage strategies. The novelty of this work lies in its comprehensive integration of multiple variables and the use of advanced machine learning optimization techniques, going beyond previous efforts by achieving higher predictive accuracy and providing a more detailed understanding of wettability dynamics.